Cationic Imidazolium-Urethane-Based Poly(Ionic Liquids) Membranes for Enhanced CO₂/CH₄ Separation: Synthesis, Characterization, and Performance Evaluation

Abstract

The escalating emissions of CO₂ into the atmosphere require the urgent development of technologies aimed at mitigating environmental impacts. Among these, aqueous amine solutions and polymeric membranes, such as cellulose acetate and polyimide are commercial technologies requiring improvement or substitution to enhance the economic and energetic efficiency of CO₂ separation processes. Ionic liquids and poly(ionic liquids) (PILs) are candidates to replace conventional CO₂ separation technologies. PILs are a class of materials capable of combining the favorable gas affinity exhibited by ionic liquids (ILs) with the processability inherent in polymeric materials. In this context, the synthesis of the IL GLYMIM[Cl] was performed, followed by ion exchange processes to achieve GLYMIM variants with diverse counter anions (NTf₂⁻, PF₆⁻, and BF₄). Subsequently, PIL membranes were fabricated from these tailored ILs and subjected to characterization, employing techniques such as SEC, FTIR, DSC, TGA, DMA, FEG-SEM, and CO₂ sorption analysis using the pressure decay method. Furthermore, permeability and ideal selectivity assessments of CO₂/CH₄ mixture were performed to derive the diffusion and solubility coefficients for both CO₂ and CH₄. PIL membranes exhibited adequate thermal and mechanical properties. The PIL-BF₄ demonstrated CO₂ sorption capacities of 33.5 mg CO₂/g at 1 bar and 104.8 mg CO₂/g at 10 bar. Furthermore, the PIL-BF₄ membrane exhibited permeability and ideal (CO₂/CH₄) selectivity values of 41 barrer and 44, respectively, surpassing those of a commercial cellulose acetate membrane as reported in the existing literature. This study underscores the potential of PIL-based membranes as promising candidates for enhanced CO₂ capture technologies.

1. Introduction

In the context of contemporary technological progress, a noticeable increase in atmospheric CO₂ emissions originating from the utilization of fossil resources has emerged [1,2]. In response, the imperative to develop new technologies or refine existing ones to improve environmental impacts has become increasingly evident. A particularly promising avenue of investigation lies in CO₂ capture. This technological approach garners attention due to its applicability to major emission sources and its potential for the recovery and repurposing of CO₂ for other industrial processes [2,3]. Currently, diverse techniques for CO₂ capture exists, among which amine solutions and polymeric membranes stand out as exemplary materials employed for this purpose [4]. The amine-based capture method, which involves the chemical absorption of CO₂, is widely used in emission plants for handling combustion gases and in natural gas sweetening [5,6,7,8,9]. Despite its cost effectiveness, this approach presents challenges such as the generation of undesirable by-products, the formation of corrosive acids leading to the degradation of the internals of the reactor, volatility, and the high energy costs associated with CO₂ recovery, thus diminishing its attractiveness over time [10]. On the other hand, the utilization of membranes in CO₂ separation demonstrates significant potential due to their adaptability to high CO₂ concentrations and their more compact structural requirements compared to amine solutions [11]. Ionic liquids (ILs) constitute a class of organic salts composed of a cation and an anion, being an organic cation and an organic or inorganic anion [4,12,13,14]. Characterized by a melting temperature below 100 °C, ILs possess several properties expected of effective CO₂ capture materials, including low vapor pressure, non-flammability, thermal stability, low corrosiveness, minimal decomposition rates, lower recovery costs, and environmental compatibility vis-à-vis organic solvents [4,12,15,16]. ILs have attracted attention in the domain of combustion gas capture due to their enhanced CO₂ affinity relative to gas mixtures containing N₂ and CH₄. However, ILs present specific attributes that impede their immediate adoption in CO₂ capture plants, such as high production costs and complexity compared to amine solutions, as well as high viscosity resulting in suboptimal CO₂ sorption/desorption rates [4]. Polymerized ionic liquids or poly(ionic liquids) (PILs) represent a distinct class of materials with broad applicability across various scientific domains [17,18,19,20,21]. Typically composed of a polymeric structure hosting an ionic liquid unit for each polymer repeat unit, PILs offer superior CO₂ sorption/desorption kinetics, thus conferring a significant advantage over ILs [12,22,23,24,25]. Condensation polymerization stands out as a common method for PIL synthesis. Researchers, including Bernard 2019, da Luz 2021, and Morozova 2020 [26,27,28] have successfully synthesized PILs and assessed their properties. Materials such as PTMG (polytetramethylene glycol), PCD (polycarbonate diol), and PCL (polycaprolactone) serve as illustrative examples of polyols used as starting reagents along with diisocyanates for PIL synthesis [26,27,29]. In the case of polyurethane (PU)-based cationic PILs, the respective ILs in the form of diols are employed, thus incorporating the IL cation into the polymer chain as documented in the literature [26,29,30]. This study endeavors to synthesize and evaluate the CO₂/CH₄ separation potential of PU-based cationic PIL membranes. The IL GLYMIM[Cl] was synthesized, followed by ion exchange processes to obtain GLYMIM with different counter anions (NTf₂⁻, PF₆⁻ and BF₄⁻). These ILs were used as diols in the synthesis of PILs, aiming to tailor polymeric chains with high affinity for CO₂, thereby producing membranes with improved selectivity.

2. Materials and Methods

2.1. Materials

For the synthesis of hydroxyl-functionalized ionic liquids (ILs), N-methylimidazole (Sigma-Aldrich, Burlington, MA, USA), 3-chloro-1,2-propanediol (Sigma-Aldrich), ethyl acetate (Neon Química, São Paulo, Brazil), lithium bis(trifluoromethanesulfonyl)imide (LiNT₂F, Sigma-Aldrich, USA), lithium tetrafluoroborate (LiBF₄, Sigma-Aldrich, USA), and sodium hexafluorophosphate (NaPF₆, Sigma-Aldrich, USA) were used. To obtain the PILs, polycarbonate diol (PCD, Mn = 2000 g/mol, Bayer, Berlin, Germany), hexamethylene diisocyanate (HDI, 99%, Merck, Darmstadt, Germany), dibutyltin dilaurate (DBTDL, Miracema Nuodex, Campinas, Brazil), methyl ethylketone (MEK, 99%, Mallinckrodt, Hazelwood, MO, USA) and the obtained ILs glyceryl-N-methylimidazolium chloride [GLYMIM][Cl], and derivatives ([GLYMIM][Cl], [GLYMIM][NT₂F], ([GLYMIM][BF₄] and ([GLYMIM][PF₆]) were used.

2.2. Synthesis of Hydroxyl-Functionalized Ionic Liquids

The synthesis of glyceryl-N-methylimidazolium chloride was conducted as described in the literature [25,31,32]. The reaction was carried out continuously in a glycerin bath with constant magnetic stirring at a temperature of 70 °C. During the first hour of the reaction, 10 mL (0.11 mol) of 3-chloro-1,2-propanediol was slowly added dropwise to a solution containing 12 mL (0.15 mol) of N-methylimidazole. The reaction conditions were maintained for 72 h.

Subsequently, the reaction mixture was removed from the heat and washed six times with ethyl acetate. The resulting material was then placed in a glycerin bath to maintain a constant temperature of 70 °C and dried under vacuum for 24 h.

The product was stored under a nitrogen atmosphere, resulting in a yellow oil [25,31,32]. Ion exchanges were performed in acetonitrile by reacting [GLYMIM]Cl with the respective salts: LiNT₂F to obtain [GLYMIM]NT₂F, LiBF₄ to form [GLYMIM]BF₄, and NaPF₆ to form [GLYMIM]PF₆. It should be noted that the reagents used in the ion exchange process ([GLYMIM]Cl, LiBF₄, LiNT₂F, and NaPF₆) are soluble in acetonitrile, while the by-products, LiCl and NaCl, are not soluble. This allowed for easy separation of the by-products through simple filtration [25,32,33,34]. The obtained structures can be observed in Figure 1.

AgNO₃ tests were carried out until no significant amount of white precipitate was observed, confirming satisfactory removal of the by-product. Vacuum drying was then performed to remove any excess solvent and moisture.

2.3. Cationic Poly(Ionic Liquids)

Initially, under constant stirring, the polyol PCD (0.04 mol) was melted in a five-neck reactor, the DBTDL catalyst (0.1% by weight) was added, and 50 mL of methyl ethyl ketone was added. After reaching the reaction temperature (70 °C), the diisocyanate (0.047 mol) was added to form the prepolymer. Then, dialcohol ([GLYMIM]Cl) (0.11 mol) and diisocyanate (0.11 mol) were added to form the PIL. The completion of the reaction is indicated by the disappearance of the free NCO band (around 2270 cm⁻¹) in the infrared spectrum. The synthesized PILs were labeled as PLIX, where X represents the counteranion (Cl, NT₂F, PF₆, or BF₄). Figure 2 shows the structural formula of a cationic PIL.

2.4. Syntheses of PU

The PU preparation process was the same as the method used for the synthesis of cationic poly(ionic liquid); however, [GLYMIM]X, was not added, using an NCO/OH molar ratio of 1.1.

2.5. Preparation of Dense Membranes

Membranes were prepared using the casting or solvent evaporation method with thickness between 150 and 250 μm. Initially, a 20% w/w solution of PU or PIL was prepared by dissolving 2 g of PIL in 10 mL of methyl ethylketone by magnetic stirring and heating (50 °C) until the PIL was completely dissolved. The solution was then applied to a flat surface, such as a glass plate or a Petri dish, followed by solvent evaporation at room temperature for five days.

2.6. Characterization

Hydroxyl-functionalized ILs and synthesized PILs were characterized by Fourier transform infrared spectroscopy (FTIR, PerkinElmer Spectrum 100 spectrometer, PerkinElmer, Waltham, MA, USA) in transmission mode in the range of 4000–650 cm⁻¹ to verify their structures. The synthesized ILs were also characterized by nuclear magnetic resonance (NMR) using a Bruker Ascend 400 model NMR spectrometer, Bruker, Billerica, MA, USA.

The molar masses of the PIL membranes and their distribution were determined by gel permeation chromatography (GPC), with chromatograms obtained using the isocratic HPLC pump-1515 chromatograph (LabMakelaar Benelux, Zevenhuizen, The Netherlands) coupled with a Waters Instruments 2412 (Waters Corp, Milford, CT, USA) refractive index detector and THF as eluent.

Scanning electron microscopy with field emission (FEI Inspect F50, FEI Company, Hillsboro, OR, USA) analyses were performed in secondary electrons (SE) mode to evaluate the surface of the composite membranes.

Differential scanning calorimetry (DSC, TA Instruments Q20, TA Instruments, New Castle, DE, USA) was used to determine the glass transition temperature (Tg), melting temperature (Tm), and crystallization temperature (Tc) of the PILs. Tests were performed with two heating ramps and one cooling ramp in the range of −90 to 200 °C at 10 °C/min under an inert nitrogen atmosphere.

The thermal stability of the PILs was assessed through thermogravimetric analysis (TGA) using a TA Instruments Q600 model instrument, ranging from room temperature to 600 °C with a heating rate of 10 °C/min under a nitrogen atmosphere.

Mechanical analyses were performed in triplicate according to ASTM D822 standard [35] on a TA Instruments Q800 (TA Instruments, New Castle, DE, USA) apparatus to determine Young’s moduli and obtain stress–strain curves.

2.7. CO₂ Sorption Capacity

The pressure decay technique was used to determine the CO₂ sorption capacity. A detailed description of the sorption apparatus and measurement procedure can be found in our previous works [26,27]. Samples (1.0 g) were placed in the sorption chamber and degassed under vacuum (10⁻³ mbar) for 1 h at room temperature before the test began. CO₂ sorption experiments were carried out at 30 °C at different equilibrium pressures (1 bar and 10 bar).

2.8. CO₂ Permeability and Ideal CO₂/CH₄ Selectivity

The permeation and selectivity for CO₂ and CH₄ of the PLI membranes were evaluated in a system consisting of two plates joined in a 4 cm diameter hole (Figure S1), where the membrane is inserted. Vacuum was applied to the membrane and the system before gas feed. CO₂ or CH₄ was fed at a pressure of 4 bar. At the bottom, a pressure transducer computed the amount of gas passing through the membrane over time (dp/dt). The permeate gas was assumed to exhibit ideal behavior; therefore, normal temperature and pressure conditions were used. Permeability was determined from the slope (dp/dt) of the linear portion of pressure versus time using Equation (1):

$$\frac{P}{l}=\frac{dp}{dt}\left(\frac{V_{System}}{A ∆p}\right)\left(\frac{T_{CNTP}}{T P_{CNTP}}\right)$$

(1)

where l is the membrane thickness, P is permeability, Δp is the pressure difference across the membrane, A corresponds to the membrane area, V_System is the volume of the permeation cell, T is the ambient temperature, and T_CNTP and P_CNTP are, respectively, the temperature and pressure at normal conditions [36]. The ideal selectivity was calculated from Equation (2) by dividing the CO₂ permeability by the CH₄ permeability.

$${\propto }_{{CO}_2/{CH}_4}=\frac{P_{{CO}_2}}{P_{{CH}_4}}$$

(2)

The solution-diffusion mechanism is widely accepted as the primary transport mechanism for gas permeation through a dense membrane [37,38]. In this mechanism, the gas solubility coefficient (cm³(STP)/(cm³ cmHg)) is calculated using Equation (3):

$$S=\frac{P}{D}$$

(3)

where P represents permeability and D represents the gas diffusion coefficient (cm²/s). The gas diffusion coefficient was determined using the time-lag method described by Equation (4) [37,39]:

$$D=\frac{l^2}{6\theta }$$

(4)

where D is the diffusion coefficient (cm²/s), l is the membrane thickness (cm), and θ is the diffusion time lag (s) determined by the linear portion of the curve intercepting the time axis.

3. Results

3.1. Characterization of the ILs

Figure S2 shows the FTIR spectra of the ILs. The IL [GLYMIM]Cl exhibited the expected bands confirming the desired product formation. The OH band between 3281 and 3239 cm⁻¹ was visible, as well as bands related to the imidazolium ring near 1670 cm⁻¹ (C=C), 1604 cm⁻¹ (N-H), and 1371 cm⁻¹ (aromatic C-N) [25,40]. The substitution of the Cl anion by NT₂F resulted in the appearance of bands between 1200 cm⁻¹ and 1400 cm⁻¹, characteristic for this ion, as well as others near 1060 cm⁻¹ corresponding to S=O bonds, 846 cm⁻¹ corresponding to N-S bonds, 789 cm⁻¹ corresponding to C-S bonds, and 751 cm⁻¹ corresponding to C-F bonds. Substitution of Cl by BF₄ and PF₆ anions also caused changes in the spectra, resulting in the appearance of bands at 815 cm⁻¹ for PF₆ P-F bonds and at 1050 cm⁻¹ for BF₄ B-F bonds [41,42,43].

For illustrative purposes, the NMR spectra of the chloride-functionalized N-glyceryl-N-methylimidazole [GLYMIM]Cl (Figure S3) and [GLYMIM]NT₂F (Figure S4) ILs, obtained after the exchange of the chloride anion (Cl⁻) for NT₂F, are shown. The 1H NMR spectrum (DMSO-d6) of [GLYMIM]Cl (Figure S2) exhibited characteristic peaks: 3.4 ppm (2H, CH2), 3.7 ppm (1H, CH), 3.9 ppm (3H, CH3), 4.4 ppm (2H, CH2), 5.3 ppm (1H, OH), 5.6 ppm (1H, OH), 7.7 ppm (1H, CH), 7.8 ppm (1H, CH), and 9.2 ppm (1H, CH) [25,32,44]. The 13C spectrum (DMSO-d6) of [GLYMIM]NT₂F (Figure S3) showed characteristic peaks: δ 36.1 (CH₃); 52.5 (-N+CH₂); 63.1 (CH₂OH); 70.07 (CHOH); 123.5 (CH); 123.6 (CH); 137.5 (CH); and 117.4 (2C, NT₂F) [25,32,44].

3.2. Characterization of Cationic Poly(Ionic Liquids)

The molecular weight distributions (PD, polydispersity) and the weighted molecular weights (Mw) of the PILs are presented in Table S1. The Mw values ranged from 43,264 gmol⁻¹ to 70,655 gmol⁻¹, while the polydispersity (PD) ranged from 1.2 to 1.4. In Figure 3, spectra obtained for the PILs are shown, and Figure 4 shows the enlargements of the spectra in the regions of 3000–3700 cm⁻¹; 1600–1800 cm⁻¹; and 650–1350 cm⁻¹. FTIR spectra (Figure 3) showed characteristic bands found for polyurethanes: 2936–2871 cm⁻¹ (CH₂ and CH₃ stretching), 1536 cm⁻¹ (N–H bending groups), 1246 cm⁻¹ (vibration of C-N and C-O groups of urethane), 1041 cm⁻¹ (stretching of C–O–C groups of urethane), and 955 cm⁻¹ (stretching of C–O–C groups of polycarbonate diol) [45]. The bands characteristic of H bonds present in polyurethanes were also evaluated: the region between 3200 and 3500 cm⁻¹ corresponding to the stretching vibration of the N-H group of urethane (Figure 4A) and the region between 1700 and 1730 cm⁻¹ characteristic of bonded and non-bonded carbonyl groups, respectively (Figure 4B). Shifts in characteristic bands (N-H and C=O) evidence possible H bonds. Figure 4A shows the characteristic bands of the stretching vibration of urethane N-H groups, where the region around 3400 cm⁻¹ corresponds to non-bonded N-H groups and the band in the region of 3300 cm⁻¹ characterizes the vibration of bonded N-H groups. All PILs showed an increase in intensity in the regions of bonded and non-bonded N-H groups with the insertion of IL into the polymeric chain. Also, concerning the structure of polyurethane, Figure 4B contains an enlargement of this area between 1600 and 1800 cm⁻¹. In this range, it is possible to observe bands related to non-bonded carbonyl groups near 1730 cm⁻¹ and bonded carbonyl groups near 1700 cm⁻¹ [27]. In Figure 4B, changes in the spectrum were also noticed only when IL insertion is performed, such as the displacement of the band ~1730 cm⁻¹ to values close to ~1710 cm⁻¹. The appearance of characteristic bands of bonded carbonyl groups in the regions of 1700 cm⁻¹ for samples with Cl, PF₆ and BF₄, and 1650 cm⁻¹ for samples with NTf₂, PF₆, and BF₄ counterions was observed, indicating the formation of H bonds [26,27]. The absence of the characteristic band for the stretching vibration of the N=C=O group in the region of 2270 cm⁻¹ confirms the absence of free NCO groups in the polymeric material, indicating the formation of PU [27,29,46]. Bands related to the counter anions present in the samples can also be visualized (Figure 3). The PIL-BF4 sample presents a broad band near 1060 cm⁻¹, indicating the presence of the B-F bond [41,42,43]. The PIL-PF₆ presents a band near 835 cm⁻¹ related to P-F bonds. For PIL-NTf₂, a band near 1346 cm⁻¹ related to SO₂ bond vibrations was observed, as well as bands near 1136 cm⁻¹ and 1189 cm⁻¹ related to C-F bond vibrations [41,42,43].

3.2.1. DSC

The DSC curves obtained for the polymeric samples are presented in Figure 5. The glass transition temperatures (Tg) values of −44.7, −47.9, −43.2, and −42.7 °C were found for the samples PIL-Cl, PIL-NTf₂, PIL-PF₆, and PIL-BF₄, respectively. By comparing the values obtained for the PILs with the result obtained for pure PU (−42 °C), it is possible to perceive a decreased tendency in the Tg values for all PILs in relation to the non-ionic PU, which could mean that the addition of ILs in the polymeric chain increases the separation of the polymer microphases. This separation of microphases facilitates the mobility of the chains [26,27]. From the analysis of the thermograms, it is possible to identify the existence of an endothermic peak which can be attributed to the melting of a crystalline microphase (Tmf). This endothermic peak is characteristic for PU-based polymers that use polyol PCD with a molecular weight of 2000 g/mol in the polymer chain [26,27]. The results found for the PU, PIL-Cl, PIL-NTf₂, PIL-PF₆, and PIL-BF₄ samples were, respectively, 40.1 °C, 42.2 °C, 46.4 °C, 42.8 °C, and 43.5 °C. Finally, the crystallization temperature (Tc) was obtained for the PIL-Cl sample (−12.7 °C).

3.2.2. TGA

Table 1. Results of degradation temperatures obtained by TGA.

Sample	Tonset1 (°C)	Tonset2 (°C)
PU	276	428
PIL-Cl	279	469
PIL-NTf2	318	462
PIL-PF6	300	440
PIL-BF4	325	445

and Figure S5 present the results of the degradation temperature obtained by TGA. PIL samples showed their first degradation temperature between 279 °C and 325 °C. This initial decomposition temperature corresponds to the rigid segments, where the degradation of the polyurethane bonds occurs. The second stage occurs around 440–469 °C, which refers to the breakdown of the flexible segments, leading to the degradation of the polyol along with the imidazole molecules. The incorporation of ILs led to an increase in the decomposition temperature, which may be related to the hydrogen bonds that may occur between PU and IL. Thermal decomposition of ILs typically occurs between 200 and 350 °C, suggesting that the decomposition of ILs may be occurring in the first stage of PILs degradation [47,48,49].

3.2.3. DMA

Figure 6 shows the stress–strain curves referring to the PILs samples. In terms of Young’s modulus, the increasing order is PIL-NTf ₂ < PIL-Cl < PIL-PF₆ < PIL-BF₄ < PU. The Young’s modulus values obtained for the PILs samples were 7 MPa (PIL-NTf₂), 16.4 MPa (PIL-Cl), 29 MPa (PIL-PF₆), 47 MPa (PIL-BF₄), and 53 MPa (PU). The insertion of an IL into the polymeric chain can act as a plasticizer, interfering with the flexibility and mobility of the chains, affecting the Young’s modulus and resulting in its decrease [27,47]. Another possibility for this material behavior is the increased separation of the polymer microphases, which can also lead to a decrease in the Young’s modulus and elongation of the curve [50,51]. Except for PIL-NTf₂, it can be observed that, besides the decrease in Young’s modulus, which indicates greater elasticity of the material, the curve becomes more elongated when comparing the PIL samples with the non-ionic PU sample.

3.3. CO₂ Sorption Capacity

The CO₂ capture results can be observed in Figure 7. From the values found, it can be noticed that increasing pressure also increases the amount of captured CO₂. This behavior is typical of samples that undergo CO₂ sorption through physical interaction, as is the case with the PILs obtained in this work. The results obtained for the non-ionic PU sample were 24.7 mg CO₂/g (1 bar) and 83.1 mg CO₂/g (10 bar). These CO₂ capture results before the addition of the ionic liquid can be attributed to the interactions between CO₂ and the polar groups composed of nitrogen and oxygen present in the structure of the PU polymer chain. Additionally, Figure 7 shows that the insertion of the IL leads to a slight increase in CO₂ capture values for all PIL samples. The results obtained for 1 bar were 27.5 mg CO₂/g for PIL-PF₆, 26.3 mg CO₂/g for PIL-Cl, 27.6 mg CO₂/g for PIL-NTf₂, and 33.5 mg CO₂/g for PIL-BF₄. For the same samples at 10 bar, the results were 93.7 mg CO₂/g for PLI-PC-Cl, 95.4 mg CO₂/g for PLI-PC-PF6, 98.1 mg CO₂/g for PIL-NTf₂, and 104.8 mg CO₂/g for PIL-BF_4. The sorption capacity increases in the following order: [Cl]⁻ < [PF₆]⁻ < [NTf₂]⁻ < [BF₄]⁻ for 1 and 10 bar. As expected, fluorinated anions showed a higher affinity for CO₂. However, in this case, the size of the anion also appears to influence the CO₂ sorption capacity. The structures of the NTF₂⁻ and PF₆⁻ anions are larger when compared to the BF₄⁻ [41,52]. This may block some important CO₂ interaction sites in the PIL cation, reducing the sorption capacity [53]. In the tests conducted, the CO₂ sorption capacity of PIL-BF₄ was superior compared to PU-based PILs reported in the literature using the BF₄ anion. For example, Morozova 2017 [29] and Morozova 2020 [28] obtained 24.7 mgCO₂/g for the PIL-8.1.BF₄ and 24.8 mgCO₂/g for PUR2.BF₄ at 1 bar and 273 K.

3.4. Permeability and CO₂ Selectivity

Table 2. Permeability and ideal selectivity results for cationic PILs (25 °C and 4 bar), compared with results for cellulose acetate and PILs from the literature.

Membrane	Perm. CO2 (Barrer)	Perm. CH4 (Barrer)	Sel. Ideal CO2/CH4
PU	2.0 ± 0.1	0.78 ± 0.02	2.5
PIL-Cl	17 ± 2.2	0.6 ± 0.1	30
PIL-NTf2	-	-	-
PIL-PF6	9 ± 1.7	0.6 ± 0.2	14
PIL-BF4	41 ± 3.2	0.9 ± 0.3	44
1—CTA (25 °C, 5 bar)	18	1.98	9
2—CA (25 °C, 3 bar)	15	1.45	10
3—CA (25 °C, 1 bar)	4	0.2	17
4—CA (35° C, 3 bar)	4.3	0.21	21
5—Styrene-Based Poly(RTILS) (2 atm; RT)	9.2 ± 0.5	0.24 ± 0.01	39
5—Acrylate-Based Poly(RTILS) (2 atm; RT)	7.0 ± 0.4	0.19 ± 0.02	37
6—PIL-TFSI (1 atm; RT)	4.1 ± 0.1	-	41
7—OEG1 (2 atm; 295 K)	16 ± 1	0.48 ± 0.01	33
7—OEG2 (2 atm; 295 K)	22 ± 1	0.74 ± 0.02	29
8—PIL-i-propyl (3 atm; 20 °C)	10.4 ± 0.2	0.35 ± 0.01	30
9—Poly[DPyDBzPBI-BuI][Tf2N] (20 atm; 35 °C)	36.2	1.25	29

1—Raza, A. et al., 2021 [54]; 2—Mubashir, M. et al., 2018 [55]; 3—Saijan, P. et al., 2020 [57]; 4—Akbarzadeh, E. et al., 2021 [58]; 5—Bara, J. et al., 2007 [59]; 6—Vollas, A. et al., 2018 [41]; 7—Bara, J. et al., 2008 [60]; 8—Horne, W. et al., 2015 [61]; 9—Shaligram, S. et al., 2016 [62].

presents the results of the CO₂ permeability and ideal (CO₂/CH₄) selectivity obtained for dense membranes of cationic PILs and pristine PU, compared to the results of cellulose acetate and some PILs found in the literature. The incorporation of ILs into the polymer chains produced membranes with superior performance when compared to neat PU, as seen in Table 2. It is observed that PIL-CL and PIL-BF₄ samples tested exhibited higher CO₂/CH₄ selectivity than cellulose acetate [54,55]. Furthermore, it can be highlighted that the PIL-BF₄ sample showed the best permeability and selectivity results among all samples presented in Table 2. The PIL-NTf₂ sample presents poor mechanical properties, breaking during the test. Better CO₂ permeability is expected for PILs, as the diffusion and solubility of CO₂ molecules are generally high in ionic liquids [56]. The influence of PIL anions on CO₂ affinity is well documented in the literature [53] and corroborated in our experiments. Permeability and selectivity increased when the Cl⁻ anion was replaced by BF₄⁻, indicating that the presence of fluorine in PILs can enhance CO₂ affinity. However, interestingly, the PF₆⁻ anion showed lower permeability and selectivity compared to the other anions. This behavior may be related to the size of the anion. The sizes of the anions used in this work follow the general trend Cl⁻ < BF₄⁻ < PF₆⁻ <NTF₂⁻ [41,52]. Therefore, although BF₄⁻ and PF₆⁻ have similar structures, PF₆ has a larger size. Bulky anion structures can reduce the free volume of the PIL, hindering CO₂ penetration towards the cation, which is primarily responsible for sorption [53]. According to Vollas et al. [41], the presence of smaller anions such as BF4⁻ facilitates polymer packing through chain interactions, leading to denser structures and consequently lower methane permeabilities. Therefore, there appears to be an optimal size for fluorinated anions (BF₄⁻) to achieve high permeability and selectivity.

To obtain more information about the gas transport mechanism through the membranes, the diffusion and solubility coefficients for CO₂ were determined. These coefficients can provide a better understanding of the variations in permeability obtained for the different PILs, as presented in Table 2. Furthermore, these coefficients can offer a better understanding of the permeability variations of PIL membranes with different counter anions, as presented in

Table 3. Diffusion and solubility coefficients for PIL membranes at 4 bar and 25 °C.

Membrane	D (10−8 cm²/s)		S (10−2 cm³(STP)/(cm³ cmHg))		DCO2/CH4	SCO2/CH4
	CO2	CH4	CO2	CH4
PIL-Cl	21.60	2.32	0.78	0.26	9.32	3.04
PIL-NTf2	-	-	-	-	-	-
PIL-PF6	22.04	1.61	0.41	0.37	13.68	1.10
PIL-BF4	32.67	1.79	1.26	0.50	18.22	2.52
CA (35 °C, 3 bar) [57]	2.05	0.55	2.11	0.38	3.72	5.55

. The values of the diffusion and solubility coefficients showed that the obtained PIL membranes exhibited higher diffusivity and solubility to CO₂ than to CH₄, which contributed to the achieved CO₂ permeabilities. Additionally, this behavior may be related to the high affinity of ILs with the polar groups of CO₂. It can be highlighted that the PIL-BF₄ sample showed higher diffusion and solubility coefficients, corroborating with the sorption capacity results. It is known that CO₂ solubility is influenced by various factors, such as the size of the anions, which can lead to an increase in the free volume of PIL, thus enhancing gas solubility since there is more space available for sorption [63].

3.5. Comparison with Robeson Upper Bound

The separation performance of dense cationic PIL membranes in this study was compared with Robeson’s curves for CO_2/CH₄ [64,65], as shown in Figure 8. It can be observed that the membranes achieved performance below the Robeson limit; however, the PIL-BF₄ sample is better positioned when compared to samples with Cl⁻ and PF₆⁻ anions, being very close to the upper limit of 1991. A likely explanation lies in the relatively low CO₂ permeability of the PILs with Cl⁻ and PF₆⁻ anions, although they show superior results compared to commercial polymer cellulose acetate. The highlight of the PIL-BF₄ sample is the ideal selectivity for CO₂/CH₄, which is higher than in other studies reported in the literature [66].

In this context, the obtained permeability and ideal selectivity results demonstrate that the synthesized PIL membranes are extremely promising for CO₂/CH₄ separation, especially because there is an intention to explore structural modification of the polymer chain to enhance the permeability and selectivity of these membranes [41]. It can be observed that the samples of cationic PILs prepared with different counter anions exhibited good permeability and high selectivity for CO₂. From the SEM images shown in Figure 9, it can be seen that the PIL membranes produced for the permeability tests exhibited a surface without pores and defects.

4. Conclusions

In this study, we successfully developed PIL membranes that exhibited adequate thermal and mechanical properties, high CO₂ permeability, and ideal CO_2/CH₄ selectivity. The IL GLYMIM [Cl] was synthesized and thoroughly characterized, and different counter anions’ chemical structures (NTf₂, PF₆, and BF₄) were evaluated to identify the optimal combination for physicochemical properties, sorption, and permeability. FTIR analysis confirmed the presence of the characteristic bands for both ILs and PILs. The incorporation of IL into the polymer chain resulted in a decrease in the Tg values, indicating improved membrane flexibility. Among the samples tested, PIL-BF₄ demonstrated the best performance in terms of sorption, permeability, and ideal CO₂/CH₄ selectivity. This suggests that appropriately sized fluorinated anions significantly enhance CO₂ permeability and ideal CO_2/CH₄ selectivity, outperforming the commercial cellulose acetate membrane.